v3.9.0

[3.9.0] - 2026-08-11

Added

• Added support for using multiple glossaries in text and document translation
  via the glossary_ids option (up to 5 glossary IDs) in translate and
  document upload.
• Added support for using a style rule in document translation via the
  style_id option (the style_rule option) in document upload.
• Added support for using a translation memory in document translation via the
  translation_memory_id option (the translation_memory and
  translation_memory_threshold options) in document upload.
• Added support for translation memory management via DeepL.translation_memories:
  find(), segments(), destroy(), create_import(), upload_file(),
  create_export(), find_job(), wait_until_job_done(), download_export(),
  and the convenience methods import_from_filepath() and export_to_filepath().
  Every method taking a translation memory accepts a string ID or a
  TranslationMemory object. wait_until_job_done(), import_from_filepath()
  and export_to_filepath() accept an optional timeout_s to stop waiting for
  a job that does not finish. Please refer to the README for usage instructions.
• Added the creation_time and updated_time attributes to TranslationMemory.

Changed

• Reworked the test suite to run against the deepl-mock server with OpenAPI
  request and response validation enabled in CI, replacing recorded VCR
  cassettes. Added live integration and error-path specs across all endpoints,
  with resource-level unit specs for response field mapping.

Removed

• Dropped VCR cassette playback and the vcr and webmock development
  dependencies.